Award AccountsThe Chemical Society of Japan Award for Creative Work for 2012

Nonadiabatic Electronic Dynamics in Isolated Moleculesand in Solution Studied by Ultrafast TimeshyEnergy Mappingof Photoelectron Distributions

Toshinori Suzuki123

1Department of Chemistry Graduate School of Science Kyoto University Kyoto 606-8502

2CREST Japan Science and Technology Agency Tokyo 102-0076

3RIKEN Center for Advanced Photonics RIKEN Wako Saitama 351-0198

Received October 4 2013 E-mail suzukikuchemkyoto-uacjp

Nonadiabatic electronic dynamics contribute to the diversity of chemical reactions We investigate nonadiabaticdynamics in isolated molecules and aqueous solutions by time-resolved photoelectron spectroscopy For isolatedmolecules in the gas phase we combine two-dimensional imaging detection of electrons and a sub-20 fs deep UV andvacuum UV light source to measure time-evolution of the photoelectron kinetic energy and angular distributions Timeshyenergy mapping of photoelectron angular anisotropy reveals the S2 frac14 S1 internal conversion dynamics through conicalintersection in pyrazine benzene and toluene The timeshyenergy mapping is also employed to extract the photoelectronangular distribution in the molecular frame for nitric oxide For electronic dynamics in aqueous solution we employ time-resolved photoelectron spectroscopy using a liquid beam and an electrostatic or a time-of-flight electron energy analyzerSuccessful observation of ultrafast electron-transfer reactions in aqueous solutions and measurement of electron bindingenergies of solvated species suggest promising future developments of this very young field

1 Introduction

The structures and dynamics of molecules are determined byquantum mechanics while the quantum mechanical equation ofmotions of the nuclei and electrons is not exactly solvable Bornand Oppenheimer have approximated the rigorous equationby separating it into a set of equations of the nuclei and of theelectrons which facilitated a quantum mechanical descriptionof chemistry1 Based on the approximation a chemical reactionis understood as quantum-mechanical nuclear motion on anadiabatic potential energy surface created by electronic motionsThe approximation however breaks down at critical nuclearconfigurations where multiple electronic states have similarenergies Consequently nonadiabatic transitions (or hoppings)occur between different potential energy surfaces in thevicinities of these critical configurations The nonadiabatictransitions change reaction pathways and product yieldscontributing to the diversity of chemical reactions

Nonadiabatic transitions are ubiquitous in the excited statedynamics of polyatomic molecules because polyatomic mole-cules have a large number of excited electronic states withina narrow energy range which cause (avoided) crossings ofpotential energy surfaces The conical intersection is one of themost important topographic features of the surface crossings

for nonadiabatic transitions2shy5 Figure 1 shows photochemicalreaction pathways mediated by conical intersections of thepotential energy surfaces in benzene6 The characteristic funnelshape of the conical intersection facilitates efficient drainingof a nuclear wave packet from an upper to a lower potentialenergy surface Although remarkable progress has been madein our understanding of chemical reactions in the past 30 yearsnonadiabatic dynamics of polyatomic molecules remain highlychallenging subjects of research Direct experimental observa-tion of ultrafast electronic dynamics and interplay with theo-retical calculations are indispensable in this research field

Photoelectron (photoemission) spectroscopy is an experi-mental method based on the photoelectric effect and it wasinitiated in the late 1950s to early 1960s7shy9 One of the pioneersof photoelectron spectroscopy Kai Siegbahn received theNobel Prize of Physics in 198110 Photoelectron spectroscopyinduces emission of an electron from a material using a photonbeam (such as UVor X-ray radiation) and measures the kineticenergies of photoelectrons Time-resolved photoelectron spec-troscopy (TRPES) is an advanced form of photoelectron spec-troscopy using a pair of laser pulses11shy15 The pump pulsecreates a wave packet in an electronically excited state and theprobe pulse interrogates its time-evolution by photoemissionTRPES requires the laser pulse durations to be shorter than the

copy 2014 The Chemical Society of Japan

Published on the web March 15 2014 doi101246bcsj20130272

Bull Chem Soc Jpn Vol 87 No 3 341shy354 (2014) 341

time scales of the dynamics of interest and the pulse energiessufficiently high to induce ionization within the pulse dura-tions A highly sensitive electron detection method is alsoimportant for TRPES because the number of photoelectronsgenerated per optical pulse-pair must be minimized to avoidelectrostatic repulsion between the photoelectrons whichwould otherwise alter the electron kinetic energies

Ionization has several important advantages as a probingmethod of chemical reactions First of all the final state isenergetically continuous so that ionization can be inducedfrom any part of the excited state potential energy surfaces aslong as the photon energy is sufficient Second the ionizationcontinuum is degenerate for different symmetries and spinstates of photoelectrons so that ionization is almost always anallowed transition the singlet and triplet excited states areequally observed Third high sensitivity can be obtained bycollecting photoelectrons using electromagnetic fields

We employ TRPES of gas-phase reactions to elucidate fun-damental aspects of electronic dynamics in elementary chemi-cal reactions The dynamics however are significantly alteredin solution especially in polar protic solvents such as waterWhile solvent effects have been employed for centuries tocontrol solution chemistry the coupled electronic and solvationdynamics in solution are not well understood Therefore itis highly interesting to develop TRPES of liquids to tackle thisclassic problem in chemistry Since the 1970s there has beena long history of endeavors to bring volatile liquids into highvacuum photoelectron spectrometers16 For an example intri-guing devices such as a rotating wire and a rotating trundle

wetted with a sample solution were developed for this pur-pose16 A high-speed liquid microjet reduced the difficultysince a small diameter (ca 10shy20macrm) minimizes the surfacearea from which solvents evaporate and a high speed mini-mizes the temperature drop in the transport from the nozzle tothe ionization point17 The temperature of water jet at 1mmdownstream of the nozzle is estimated to be 5shy10 degC18 Wedevoted our own efforts to develop TRPES of liquids for adecade and succeeded for the first time in 20101920

This article is organized as follows In Section 2 I describeTRPES of gas-phase reactions using two-dimensional photo-electron imaging and a novel ultrafast laser in the deep UV(DUV) and vacuum UV (VUV) regions The key technique istimeshyenergy mapping of photoelectron angular anisotropywhich provides information on electron configuration of thenonstationary state We examine ultrafast internal conversionin isolated molecules of pyrazine benzene and toluene anddiscuss their differences In Section 3 I describe extraction ofthe molecular-frame photoelectron angular distribution seen byan observer on a molecule using the timeshyenergy mapping ofphotoelectron angular anisotropy Section 4 is devoted to thenew research area of TRPES of liquids An electron-transferreaction in water and generation of a hydrated electron arediscussed Section 5 presents the summary and perspectives forfuture studies

2 Ultrafast Internal Conversion in Isolated AromaticMolecules Studied by Time-Resolved PhotoelectronImaging Using a Sub-20 fs DUVVUV Light Source

21 Experimental Technique When photoelectron spec-troscopy was initiated in the 1950s using continuous lightsources experiments were performed using electrostatic elec-tron energy analyzers7shy9 Although these analyzers are stillprincipal instruments in high-resolution X-ray photoelectronspectroscopy today relatively long integration times of thesignals are required owing to small solid angles and narrowenergy windows With the development of intense pulsed(nanosecond and picosecond) lasers photoelectron spectrosco-py using multiphoton ionization started in the 1980s whichemployed time-of-flight (TOF) photoelectron spectrometersThe TOF spectrometer measures flight times of photoelectronsfrom a sample to a detector and it enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basisFurthermore a magnetic bottle was introduced to achieve anelectron collection efficiency of 50 with the TOF analyzerwhich enabled the application of photoelectron spectroscopy tolow-density species such as molecular and metal clusters insupersonic beams21 However a photoelectron angular distri-bution can hardly be measured with a magnetic bottle as itutilizes electron cyclotron motions in a magnetic field

Chandler and Houston invented an ion imaging techniquein 1988 in which ions emitted from a small target volumein a Wiley-McLaren TOF mass spectrometer are acceleratedand projected onto a two-dimensional (2D) position-sensitivedetector22 Eppink and Parker have modified the ion opticelectrodes and enabled 2D space focusing of ions to improvethe imaging resolution23 Their method is now called velocitymap imaging because the arrival position of the ion on thedetector plane is proportional to the velocity perpendicular

Figure 1 Photochemical reactions of benzene from S2 andS1 states mediated by conical intersections Reproducedwith permission from Ref 6

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)342

to the flight axis and independent of the ionization pointThe performance of acceleration electrodes (aberration etc)can be further improved by increasing number of electrodesbeyond the EppinkshyParker minimal design which uses threeelectrodes2425

We combined the femtosecond pumpshyprobe method and2D imaging technique to initiate time-resolved photoelectronimaging (TRPEI) in 19992627 Figure 2 shows a schematicdiagram of TRPEI28shy30 An ultracold gas of target molecules iscreated by adiabatic gas expansion into vacuum and the gas jetis skimmed to create a supersonic molecular beam 2mm indiameter The beam is introduced into a photoelectron spec-trometer and crossed with the pump and probe laser beams Thepump pulse excites molecules to excited electronic states andthe probe pulse induces photoemission to create an expandingsphere of a photoelectron distribution The photoelectronsare accelerated in a static electric field and projected onto aposition-sensitive detector The detector consists of micro-channel plates a phosphor screen and a CCD (or CMOS)camera and it records the arrival positions of the photo-electrons on the detector plane Since both the pump and probelaser polarizations are parallel to each other and to the detectorface the original 3D distribution has axial symmetry aroundthe polarization direction With this symmetry the 3D distri-bution can be reconstructed from the projection image Theadvantage of TRPEI over other methodologies is the ability tomeasure time-evolution of the photoelectron angular distribu-tion (PAD) with high efficiency and accuracy We developed a2D electron-counting apparatus using a CMOS camera andreal-time centroiding calculations on a field programmable gatearray circuit which captured electron images at 1 kHz witha uniform sensitivity over the detector area31 The imagingdetector is able to record simultaneous arrivals of many elec-trons which is an advantage over other 2D position-sensitivedetectors such as delay-line detectors

The photoelectron ejection angle is an important observa-ble in photoelectron spectroscopy Since the initial ensemble

of molecules has an isotropic molecular axis distribution inthe absence of an external field it is an isotropic target forphotoionization similar to the 1s orbital of a hydrogen atomTherefore the anisotropy of the total physical system (mole-cule + radiation) prior to photoionization is caused by polar-ized photons This anisotropy is transferred to the PAD afterphotoionization For linear polarization of the pump and probepulses parallel to each other the photoelectron kinetic energyand angular distribution in [1 + 1curren] photoionization is ex-pressed as follows (the prime means different color)

Ietht E ordfTHORN frac14 middotetht ETHORN4sup3

f1thorn cent2etht ETHORNP2ethcos ordfTHORNthorn cent4etht ETHORNP4ethcos ordfTHORNg eth1THORN

where t ordf and E are the pumpshyprobe time delay the electronejection angle from the laser polarization direction and thephotoelectron kinetic energy Pn(x) are the n-th order Legendrepolynomials middot(tE) represents a photoelectron kinetic energydistribution or photoelectron spectrum cent2(tE) and cent4(tE)are called anisotropy parameters and the three scalar quan-tities middot(tE) cent2(tE) and cent4(tE) in eq 1 are the observables in[1 + 1curren] TRPEI of gaseous samples

Now we consider laser requirements in TRPEI experimentsby considering the most fundamental aromatic molecule ofbenzene Its ionization energy is 924 eV so that one-photonionization of benzene requires an ionization laser wavelengthshorter than 134 nm and the pumpshyprobe TRPES requiresat least one of the pump or probe wavelengths to be shorterthan 268 nm One of the characteristic molecular vibrations ofbenzene is a totally symmetric ring-breathing mode with avibrational period of ca 50 fs In order to create a spatiallylocalized wave packet and observe its time evolution thepump and probe pulses should be sub-30 fs While tunableDUV pulses down to ca 200 nm can be generated using opticalparametric amplifiers and nonlinear optical crystals a VUVpulse can hardly be generated by these methods The pulsedurations are also generally longer than 30 fs Thus we devel-

Figure 2 Time-resolved photoelectron imaging apparatus

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 343

oped a filamentation four-wave mixing method32shy34 in raregas to generate sub-20 fs DUV and VUV pulses (Figure 3)Filamentation is a unique propagation scheme of an intenselaser pulse through a medium An intense laser pulse altersthe refractive index of the medium (optical Kerr effect) whichinduces self-focusing of the pulse however as the focusedlaser pulse induces (tunnel-)ionization of the medium theionized species create an opposite spatial gradient of the refrac-tive index Consequently the laser pulse propagates throughthe medium as a tightly focused beam for a distance longer thanthe Rayleigh length35 which facilitates efficient wavelengthconversion in low density gases

In our experiment the fundamental (frac12 775 nm 25 fs) andthe second harmonic (2frac12 390 nm 30 fs) of a Tisapphire laserare gently focused into a neon gas inducing the filamentationfour-wave mixing process of 2frac12 + 2frac12 sup1 frac12 frac14 3frac12 (264 nm)Then the cascaded four-wave mixing processes generate 4frac12(198 nm) 5frac12 (159 nm) and 6frac12 (134 nm) pulses simultane-ously The DUV and VUV pulses are transmitted through apinhole placed at the end of the neon gas cell The central andperipheral parts of the output beam are separated spatiallyand different harmonics are isolated using dichroic mirrorsThe two different harmonics are independently focused onto amolecular beam with two concave mirrors The cross-correla-tion between 4frac12 and 5frac12 is 18 laquo 2 fs

22 S2 frac14 S1 Internal Conversion via a Conical Intersec-tion in Pyrazine Pyrazine was one of the first moleculesstudied with TRPES in 199927 The S2(1B2u ππ) frac14 S1(1B3unπ) internal conversion of pyrazine (C4H4N2 D2h) is the best-known example of ultrafast electronic deactivation mediatedby a conical intersection The low-lying conical intersectionbetween the S2 and S1 potential energy surfaces of pyrazinewas identified in 198836 Thereafter a number of theoreticalstudies have been performed on this system37shy50 Although

pyrazine has 24 normal modes only a single mode Q10a (b1g)mediates the S2shyS1 coupling and a few totally symmetric (ag)modes play principal roles in the vibrational dynamics Seeland Domcke proposed TRPES of pyrazine theoretically in1991 using model calculations taking into account three vibra-tional coordinates of Q1 Q6a and Q10a51 Theoretical studieshave predicted that the internal conversion occurs within30 fs while the time resolution of our experiment in 1999was 450 fs2627 Thus we constructed a sub-20 fs DUV laserusing filamentation four-wave mixing and revisited the S2shyS1dynamics of pyrazine in 201052

Figure 4 shows the UV photoabsorption spectra of pyrazine(pyrazine-h4) and fully deuterated pyrazine (pyrazine-d4) vapormeasured at room temperature Overlaid are the spectra of our3frac12 (264 nm blue) and 4frac12 (198 nm red) pulses The 3frac12 pulseexcites pyrazine to S2 near the origin and the 4frac12 pulse ionizesfrom S2 and S1 However the 4frac12 pulse overlaps (unfavorably)

Figure 3 (a) Schematic diagram of the experimental setup (b) Schematic drawing of the differential pumping systemAll dimensions are given in millimeters

Figure 4 UV photoabsorption spectra of S1 S2 and S3 ofpyrazine-h4 (thin solid line) and pyrazine-d4 (thin dashedline) at room temperature The spectra of our pump(264 nm 470 eV) and probe (198 nm 626 eV) pulses arealso shown in solid lines

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)344

with the S3shyS0 band and the 4frac12shy3frac12 pulse sequence in thenegative time delays also creates a photoelectron signal

Figure 5a shows the total photoelectron signal intensity asa function of the pumpshyprobe time delay For positive timedelays with the 3frac12shy4frac12 pulse sequence the total signal inten-sity rapidly decays within the first 30 fs and exhibits a plateauat later times our previous experiment in 1999 has shown thatthis long-lived component has a finite lifetime of 22 ps forpyrazine-h42753 For negative time delays with the 4frac12shy3frac12pulse sequence the signal decays within 50 fs These timeprofiles therefore consist of three components correspondingto the decay of optically excited S2 (red) the correspondinggrowth of S1 (blue) populated by internal conversion from S2and the decay of S3 (green) respectively Using least-squaresfitting the S2 frac14 S1 internal conversion time constants areestimated as 23 laquo 4 fs for pyrazine-h4 and 20 laquo 2 fs forpyrazine-d4 The plateau region exhibits oscillatory featuresdue to vibrational quantum beats of Q6a in S1 (583 cmsup11)

Figure 5b shows the photoelectron kinetic energy distribu-tions measured at each time delay Although the S2 frac14 S1internal conversion occurs within 30 fs no marked change isobserved here This is because photoionization predominantly

occurs as D0(nsup11)larr S1(nπ) and D1(πsup11)larr S2(ππ) and theenergy differences are almost identical between D1 and D0

(088 eV)54 and between S2 and S1 (086 eV)55

Figure 5c shows the time and energy dependence of theanisotropy parameter cent2(tE) the positive (blue-green) andnegative (red) values correspond to preferential ejection of anelectron parallel and perpendicular to the probe laser polar-ization (eq 1) The energy-dependence of cent2 a stripe of colorsat each time delay in Figure 5c is the fingerprint of theelectronic character In the region of t lt 30 fs high-energyelectrons are ejected perpendicularly to the probe laser polar-ization (red) while low-energy electrons are distributed almostisotropically (green) After 30 fs the high-energy electrondistribution becomes isotropic while low-energy electrons arepreferentially ejected parallel to the probe laser polarization(blue) The sudden change of the color at ca 30 fs indicates theoccurrence of S2 frac14 S1 internal conversion in agreement withthe analysis of the total electron signal

In Figure 5c the photoelectron kinetic energy component of08shy10 eV appearing within 30 fs is attributed to photoioniza-tion from S2(ππ) to D0(nsup11) As shown in Figure 6 themain electron configurations of S2(ππ) to D0(nsup11) cannot beconnected by a one-electron process The occurrence of theD0(nsup11) larr S2(ππ) ionization suggests that there are otherelectron configurations involved in the cationic andor neutralstate Within the HartreeshyFock approximation the electronicconfiguration of pyrazine in its ground state (S0) is given by

1b21u1a2g1b

22u2a

2g2b

21u1b

23g3a

2g3b

21u2b

22u4a

2g2b

23g5a

2g3b

22u4b

21u4b

22u3b

23g

1b23uethsup3THORN5b21uethnTHORN1b22gethsup3THORN1b21gethsup3THORN6a2gethnTHORN2b03uethsup3THORN1a0uethsup3THORN2b02gethsup3THORNeth2THORN

where nonbonding and π orbitals are respectively denoted byn and π and all other orbitals are σ orbitals Denoting the differ-ence from eq 2 the leading configurations of D0 and S2 areexpressed by 6a1

g and 1b11g 2b

thorn13u respectively To enable D0 larr

S2 ionization D0 andor S2 must involve minor electron con-figurations with a singly occupied b2u orbital and two-electronexcitation from the leading configuration jb1

2u 1b11g 2b

thorn13u ( j =

1 2 3 and 4) and 6a1g jbthorn1

2u ethj 5THORN are the candidates forD0 and S2 respectively as shown in Figures 6b and 6d Inorder to ascertain this speculation we performed first-orderconfiguration interaction and continuum multiple scattering

Figure 5 (a) Temporal profile of total photoelectron signalin 3frac12shy4frac12 TRPEI of pyrazine-h4 The observed profile iswell explained by three components single-exponentialdecay of S2 (red) corresponding increase in S1 (blue) forpositive time delays and single-exponential decay of S3(green) for negative time delays The fitting result is shownas a solid line (b) Time-evolution of photoelectron kineticenergy (PKE) distribution middot(tE) (c) Time evolution of thephotoelectron angular anisotropy parameter cent2(tE)

D0(Ag)

S2(B2u)

2b3u

2b3u

6ag

6ag

1b1g

1b1g

j b2u

j b2u

i b2u

Figure 6 Main configurations of (a) D0 and (c) S2 andrelevant configurations for one-photon ionization of D0 larrS2 (b) and (d)

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 345

Xα (CMSXα) calculations which indicated that the spectralintensity of D0 larr S2 is 47 from 3b2u 27 from 4b2u and26 from all virtual b2u ( jb2u j sup2 5)56

The characteristic negative cent2 associated with the D0 larr S2transition assisted identification of the S2 frac14 S1 internal con-version in pyrazine Using CMSXα calculations we found thatcent parameter diminishes with the electron kinetic energy owingto the energy-dependent Coulomb phases but cent does notbecome negative without a shape resonance caused by a short-lived bound electronic state buried in the ionization continuumOur calculations indicate that there is a shape resonance at35shy40 eV above the ionization energy56 Aromatic moleculesoften have shape resonances within the range of excess energyless than 10 eV which should be taken into account in theinterpretation of TRPEI

Figure 5b indicates that the FranckshyCondon distributionupon photoionization extends toward an unobservable negativekinetic energy region A probe photon with higher energy willenable observation of the entire FranckshyCondon envelopeThus we revisited the same reaction using the filamentationVUV light source TRPES of pyrazine using the 3frac12 pump and5frac12 probe is shown in Figure 7 In the positive time delayregion photoionization from the S2(ππ) and S1(nπ) areclearly seen in the range of 15shy25 eV Notice that the S1component has shifted to higher photoelectron kinetic energiesin Figure 7 in comparison with Figure 5b Since 5frac12 is stronglyabsorbed by pyrazine the 5frac12shy3frac12 pulse sequence also producesa signal in the negative time delay region The three peaksin the energy range of 15shy27 eV are of the 3s 3pz and 3pyRydberg states5457 The distribution in the 0shy2 eV range is dueto photodissociation from higher valence states excited with5frac12 so that the photoelectron kinetic energy diminishes rapidlyas the vibrational wave packet moves away from the FranckshyCondon region

23 S2 frac14 S1 Internal Conversion via Conical Intersectionin Benzene and Toluene Benzene is a benchmark aromaticmolecule for theoretical and experimental studies of organiccompounds The S2(1B1u) states of benzene and its derivativesare short-lived owing to ultrafast S2 frac14 S1 internal conversion

Let us examine the dynamics in comparison with the internalconversion in pyrazine

Figures 8a and 8b show the 4frac12 pump and 3frac12 probe TRPESof benzene and toluene respectively58 The broken lines aresingle exponential decay functions convoluted with the cross-correlation of the pump and probe pulses The discrepancybetween the data points and the simulated curves indicate thatthe single exponential decay model does not adequately re-produce the observed time profiles in either case The observeddecay profiles exhibit delayed responses which correspond topropagation of a wave packet from the FranckshyCondon regionto the conical intersection region The arrival times of the wavepackets at the seam of crossings are estimated as 33 fs forbenzene and 41 fs for toluene the subsequent population decaytime is 32 fs for benzene and 43 fs for toluene

Figure 7 Photoelectron kinetic energy distribution ob-served using the 3frac12 and 5frac12 pulses Positive time delayscorrespond to the case where the 3frac12 pulse precedes the5frac12 pulse

Figure 8 The time profiles of the photoionization signalintensity for (a) benzene and (b) toluene Photoelectronsignals are indicated by green dots with error bars Timeshyenergy maps of the photoelectron intensity σ(tE) for(c) benzene and (d) toluene Timeshyenergy maps of thephotoelectron angular anisotropies cent2(tE) for (e) benzeneand (f ) toluene and cent4(tE) for (g) benzene and (h) tolueneData points for cent2 and cent4 with standard deviations smallerthan 02 are shown (see text) Reproduced with permissionfrom Ref 58

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)346

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

to the flight axis and independent of the ionization pointThe performance of acceleration electrodes (aberration etc)can be further improved by increasing number of electrodesbeyond the EppinkshyParker minimal design which uses threeelectrodes2425

We combined the femtosecond pumpshyprobe method and2D imaging technique to initiate time-resolved photoelectronimaging (TRPEI) in 19992627 Figure 2 shows a schematicdiagram of TRPEI28shy30 An ultracold gas of target molecules iscreated by adiabatic gas expansion into vacuum and the gas jetis skimmed to create a supersonic molecular beam 2mm indiameter The beam is introduced into a photoelectron spec-trometer and crossed with the pump and probe laser beams Thepump pulse excites molecules to excited electronic states andthe probe pulse induces photoemission to create an expandingsphere of a photoelectron distribution The photoelectronsare accelerated in a static electric field and projected onto aposition-sensitive detector The detector consists of micro-channel plates a phosphor screen and a CCD (or CMOS)camera and it records the arrival positions of the photo-electrons on the detector plane Since both the pump and probelaser polarizations are parallel to each other and to the detectorface the original 3D distribution has axial symmetry aroundthe polarization direction With this symmetry the 3D distri-bution can be reconstructed from the projection image Theadvantage of TRPEI over other methodologies is the ability tomeasure time-evolution of the photoelectron angular distribu-tion (PAD) with high efficiency and accuracy We developed a2D electron-counting apparatus using a CMOS camera andreal-time centroiding calculations on a field programmable gatearray circuit which captured electron images at 1 kHz witha uniform sensitivity over the detector area31 The imagingdetector is able to record simultaneous arrivals of many elec-trons which is an advantage over other 2D position-sensitivedetectors such as delay-line detectors

The photoelectron ejection angle is an important observa-ble in photoelectron spectroscopy Since the initial ensemble

of molecules has an isotropic molecular axis distribution inthe absence of an external field it is an isotropic target forphotoionization similar to the 1s orbital of a hydrogen atomTherefore the anisotropy of the total physical system (mole-cule + radiation) prior to photoionization is caused by polar-ized photons This anisotropy is transferred to the PAD afterphotoionization For linear polarization of the pump and probepulses parallel to each other the photoelectron kinetic energyand angular distribution in [1 + 1curren] photoionization is ex-pressed as follows (the prime means different color)

Ietht E ordfTHORN frac14 middotetht ETHORN4sup3

f1thorn cent2etht ETHORNP2ethcos ordfTHORNthorn cent4etht ETHORNP4ethcos ordfTHORNg eth1THORN

where t ordf and E are the pumpshyprobe time delay the electronejection angle from the laser polarization direction and thephotoelectron kinetic energy Pn(x) are the n-th order Legendrepolynomials middot(tE) represents a photoelectron kinetic energydistribution or photoelectron spectrum cent2(tE) and cent4(tE)are called anisotropy parameters and the three scalar quan-tities middot(tE) cent2(tE) and cent4(tE) in eq 1 are the observables in[1 + 1curren] TRPEI of gaseous samples

Now we consider laser requirements in TRPEI experimentsby considering the most fundamental aromatic molecule ofbenzene Its ionization energy is 924 eV so that one-photonionization of benzene requires an ionization laser wavelengthshorter than 134 nm and the pumpshyprobe TRPES requiresat least one of the pump or probe wavelengths to be shorterthan 268 nm One of the characteristic molecular vibrations ofbenzene is a totally symmetric ring-breathing mode with avibrational period of ca 50 fs In order to create a spatiallylocalized wave packet and observe its time evolution thepump and probe pulses should be sub-30 fs While tunableDUV pulses down to ca 200 nm can be generated using opticalparametric amplifiers and nonlinear optical crystals a VUVpulse can hardly be generated by these methods The pulsedurations are also generally longer than 30 fs Thus we devel-

Figure 2 Time-resolved photoelectron imaging apparatus

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 343

oped a filamentation four-wave mixing method32shy34 in raregas to generate sub-20 fs DUV and VUV pulses (Figure 3)Filamentation is a unique propagation scheme of an intenselaser pulse through a medium An intense laser pulse altersthe refractive index of the medium (optical Kerr effect) whichinduces self-focusing of the pulse however as the focusedlaser pulse induces (tunnel-)ionization of the medium theionized species create an opposite spatial gradient of the refrac-tive index Consequently the laser pulse propagates throughthe medium as a tightly focused beam for a distance longer thanthe Rayleigh length35 which facilitates efficient wavelengthconversion in low density gases

In our experiment the fundamental (frac12 775 nm 25 fs) andthe second harmonic (2frac12 390 nm 30 fs) of a Tisapphire laserare gently focused into a neon gas inducing the filamentationfour-wave mixing process of 2frac12 + 2frac12 sup1 frac12 frac14 3frac12 (264 nm)Then the cascaded four-wave mixing processes generate 4frac12(198 nm) 5frac12 (159 nm) and 6frac12 (134 nm) pulses simultane-ously The DUV and VUV pulses are transmitted through apinhole placed at the end of the neon gas cell The central andperipheral parts of the output beam are separated spatiallyand different harmonics are isolated using dichroic mirrorsThe two different harmonics are independently focused onto amolecular beam with two concave mirrors The cross-correla-tion between 4frac12 and 5frac12 is 18 laquo 2 fs

22 S2 frac14 S1 Internal Conversion via a Conical Intersec-tion in Pyrazine Pyrazine was one of the first moleculesstudied with TRPES in 199927 The S2(1B2u ππ) frac14 S1(1B3unπ) internal conversion of pyrazine (C4H4N2 D2h) is the best-known example of ultrafast electronic deactivation mediatedby a conical intersection The low-lying conical intersectionbetween the S2 and S1 potential energy surfaces of pyrazinewas identified in 198836 Thereafter a number of theoreticalstudies have been performed on this system37shy50 Although

pyrazine has 24 normal modes only a single mode Q10a (b1g)mediates the S2shyS1 coupling and a few totally symmetric (ag)modes play principal roles in the vibrational dynamics Seeland Domcke proposed TRPES of pyrazine theoretically in1991 using model calculations taking into account three vibra-tional coordinates of Q1 Q6a and Q10a51 Theoretical studieshave predicted that the internal conversion occurs within30 fs while the time resolution of our experiment in 1999was 450 fs2627 Thus we constructed a sub-20 fs DUV laserusing filamentation four-wave mixing and revisited the S2shyS1dynamics of pyrazine in 201052

Figure 4 shows the UV photoabsorption spectra of pyrazine(pyrazine-h4) and fully deuterated pyrazine (pyrazine-d4) vapormeasured at room temperature Overlaid are the spectra of our3frac12 (264 nm blue) and 4frac12 (198 nm red) pulses The 3frac12 pulseexcites pyrazine to S2 near the origin and the 4frac12 pulse ionizesfrom S2 and S1 However the 4frac12 pulse overlaps (unfavorably)

Figure 3 (a) Schematic diagram of the experimental setup (b) Schematic drawing of the differential pumping systemAll dimensions are given in millimeters

Figure 4 UV photoabsorption spectra of S1 S2 and S3 ofpyrazine-h4 (thin solid line) and pyrazine-d4 (thin dashedline) at room temperature The spectra of our pump(264 nm 470 eV) and probe (198 nm 626 eV) pulses arealso shown in solid lines

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)344

with the S3shyS0 band and the 4frac12shy3frac12 pulse sequence in thenegative time delays also creates a photoelectron signal

Figure 5a shows the total photoelectron signal intensity asa function of the pumpshyprobe time delay For positive timedelays with the 3frac12shy4frac12 pulse sequence the total signal inten-sity rapidly decays within the first 30 fs and exhibits a plateauat later times our previous experiment in 1999 has shown thatthis long-lived component has a finite lifetime of 22 ps forpyrazine-h42753 For negative time delays with the 4frac12shy3frac12pulse sequence the signal decays within 50 fs These timeprofiles therefore consist of three components correspondingto the decay of optically excited S2 (red) the correspondinggrowth of S1 (blue) populated by internal conversion from S2and the decay of S3 (green) respectively Using least-squaresfitting the S2 frac14 S1 internal conversion time constants areestimated as 23 laquo 4 fs for pyrazine-h4 and 20 laquo 2 fs forpyrazine-d4 The plateau region exhibits oscillatory featuresdue to vibrational quantum beats of Q6a in S1 (583 cmsup11)

Figure 5b shows the photoelectron kinetic energy distribu-tions measured at each time delay Although the S2 frac14 S1internal conversion occurs within 30 fs no marked change isobserved here This is because photoionization predominantly

occurs as D0(nsup11)larr S1(nπ) and D1(πsup11)larr S2(ππ) and theenergy differences are almost identical between D1 and D0

(088 eV)54 and between S2 and S1 (086 eV)55

Figure 5c shows the time and energy dependence of theanisotropy parameter cent2(tE) the positive (blue-green) andnegative (red) values correspond to preferential ejection of anelectron parallel and perpendicular to the probe laser polar-ization (eq 1) The energy-dependence of cent2 a stripe of colorsat each time delay in Figure 5c is the fingerprint of theelectronic character In the region of t lt 30 fs high-energyelectrons are ejected perpendicularly to the probe laser polar-ization (red) while low-energy electrons are distributed almostisotropically (green) After 30 fs the high-energy electrondistribution becomes isotropic while low-energy electrons arepreferentially ejected parallel to the probe laser polarization(blue) The sudden change of the color at ca 30 fs indicates theoccurrence of S2 frac14 S1 internal conversion in agreement withthe analysis of the total electron signal

In Figure 5c the photoelectron kinetic energy component of08shy10 eV appearing within 30 fs is attributed to photoioniza-tion from S2(ππ) to D0(nsup11) As shown in Figure 6 themain electron configurations of S2(ππ) to D0(nsup11) cannot beconnected by a one-electron process The occurrence of theD0(nsup11) larr S2(ππ) ionization suggests that there are otherelectron configurations involved in the cationic andor neutralstate Within the HartreeshyFock approximation the electronicconfiguration of pyrazine in its ground state (S0) is given by

1b21u1a2g1b

22u2a

2g2b

21u1b

23g3a

2g3b

21u2b

22u4a

2g2b

23g5a

2g3b

22u4b

21u4b

22u3b

23g

1b23uethsup3THORN5b21uethnTHORN1b22gethsup3THORN1b21gethsup3THORN6a2gethnTHORN2b03uethsup3THORN1a0uethsup3THORN2b02gethsup3THORNeth2THORN

where nonbonding and π orbitals are respectively denoted byn and π and all other orbitals are σ orbitals Denoting the differ-ence from eq 2 the leading configurations of D0 and S2 areexpressed by 6a1

g and 1b11g 2b

thorn13u respectively To enable D0 larr

S2 ionization D0 andor S2 must involve minor electron con-figurations with a singly occupied b2u orbital and two-electronexcitation from the leading configuration jb1

2u 1b11g 2b

thorn13u ( j =

1 2 3 and 4) and 6a1g jbthorn1

2u ethj 5THORN are the candidates forD0 and S2 respectively as shown in Figures 6b and 6d Inorder to ascertain this speculation we performed first-orderconfiguration interaction and continuum multiple scattering

Figure 5 (a) Temporal profile of total photoelectron signalin 3frac12shy4frac12 TRPEI of pyrazine-h4 The observed profile iswell explained by three components single-exponentialdecay of S2 (red) corresponding increase in S1 (blue) forpositive time delays and single-exponential decay of S3(green) for negative time delays The fitting result is shownas a solid line (b) Time-evolution of photoelectron kineticenergy (PKE) distribution middot(tE) (c) Time evolution of thephotoelectron angular anisotropy parameter cent2(tE)

D0(Ag)

S2(B2u)

2b3u

2b3u

6ag

6ag

1b1g

1b1g

j b2u

j b2u

i b2u

Figure 6 Main configurations of (a) D0 and (c) S2 andrelevant configurations for one-photon ionization of D0 larrS2 (b) and (d)

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 345

Xα (CMSXα) calculations which indicated that the spectralintensity of D0 larr S2 is 47 from 3b2u 27 from 4b2u and26 from all virtual b2u ( jb2u j sup2 5)56

The characteristic negative cent2 associated with the D0 larr S2transition assisted identification of the S2 frac14 S1 internal con-version in pyrazine Using CMSXα calculations we found thatcent parameter diminishes with the electron kinetic energy owingto the energy-dependent Coulomb phases but cent does notbecome negative without a shape resonance caused by a short-lived bound electronic state buried in the ionization continuumOur calculations indicate that there is a shape resonance at35shy40 eV above the ionization energy56 Aromatic moleculesoften have shape resonances within the range of excess energyless than 10 eV which should be taken into account in theinterpretation of TRPEI

Figure 5b indicates that the FranckshyCondon distributionupon photoionization extends toward an unobservable negativekinetic energy region A probe photon with higher energy willenable observation of the entire FranckshyCondon envelopeThus we revisited the same reaction using the filamentationVUV light source TRPES of pyrazine using the 3frac12 pump and5frac12 probe is shown in Figure 7 In the positive time delayregion photoionization from the S2(ππ) and S1(nπ) areclearly seen in the range of 15shy25 eV Notice that the S1component has shifted to higher photoelectron kinetic energiesin Figure 7 in comparison with Figure 5b Since 5frac12 is stronglyabsorbed by pyrazine the 5frac12shy3frac12 pulse sequence also producesa signal in the negative time delay region The three peaksin the energy range of 15shy27 eV are of the 3s 3pz and 3pyRydberg states5457 The distribution in the 0shy2 eV range is dueto photodissociation from higher valence states excited with5frac12 so that the photoelectron kinetic energy diminishes rapidlyas the vibrational wave packet moves away from the FranckshyCondon region

23 S2 frac14 S1 Internal Conversion via Conical Intersectionin Benzene and Toluene Benzene is a benchmark aromaticmolecule for theoretical and experimental studies of organiccompounds The S2(1B1u) states of benzene and its derivativesare short-lived owing to ultrafast S2 frac14 S1 internal conversion

Let us examine the dynamics in comparison with the internalconversion in pyrazine

Figures 8a and 8b show the 4frac12 pump and 3frac12 probe TRPESof benzene and toluene respectively58 The broken lines aresingle exponential decay functions convoluted with the cross-correlation of the pump and probe pulses The discrepancybetween the data points and the simulated curves indicate thatthe single exponential decay model does not adequately re-produce the observed time profiles in either case The observeddecay profiles exhibit delayed responses which correspond topropagation of a wave packet from the FranckshyCondon regionto the conical intersection region The arrival times of the wavepackets at the seam of crossings are estimated as 33 fs forbenzene and 41 fs for toluene the subsequent population decaytime is 32 fs for benzene and 43 fs for toluene

Figure 7 Photoelectron kinetic energy distribution ob-served using the 3frac12 and 5frac12 pulses Positive time delayscorrespond to the case where the 3frac12 pulse precedes the5frac12 pulse

Figure 8 The time profiles of the photoionization signalintensity for (a) benzene and (b) toluene Photoelectronsignals are indicated by green dots with error bars Timeshyenergy maps of the photoelectron intensity σ(tE) for(c) benzene and (d) toluene Timeshyenergy maps of thephotoelectron angular anisotropies cent2(tE) for (e) benzeneand (f ) toluene and cent4(tE) for (g) benzene and (h) tolueneData points for cent2 and cent4 with standard deviations smallerthan 02 are shown (see text) Reproduced with permissionfrom Ref 58

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)346

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

oped a filamentation four-wave mixing method32shy34 in raregas to generate sub-20 fs DUV and VUV pulses (Figure 3)Filamentation is a unique propagation scheme of an intenselaser pulse through a medium An intense laser pulse altersthe refractive index of the medium (optical Kerr effect) whichinduces self-focusing of the pulse however as the focusedlaser pulse induces (tunnel-)ionization of the medium theionized species create an opposite spatial gradient of the refrac-tive index Consequently the laser pulse propagates throughthe medium as a tightly focused beam for a distance longer thanthe Rayleigh length35 which facilitates efficient wavelengthconversion in low density gases

In our experiment the fundamental (frac12 775 nm 25 fs) andthe second harmonic (2frac12 390 nm 30 fs) of a Tisapphire laserare gently focused into a neon gas inducing the filamentationfour-wave mixing process of 2frac12 + 2frac12 sup1 frac12 frac14 3frac12 (264 nm)Then the cascaded four-wave mixing processes generate 4frac12(198 nm) 5frac12 (159 nm) and 6frac12 (134 nm) pulses simultane-ously The DUV and VUV pulses are transmitted through apinhole placed at the end of the neon gas cell The central andperipheral parts of the output beam are separated spatiallyand different harmonics are isolated using dichroic mirrorsThe two different harmonics are independently focused onto amolecular beam with two concave mirrors The cross-correla-tion between 4frac12 and 5frac12 is 18 laquo 2 fs

22 S2 frac14 S1 Internal Conversion via a Conical Intersec-tion in Pyrazine Pyrazine was one of the first moleculesstudied with TRPES in 199927 The S2(1B2u ππ) frac14 S1(1B3unπ) internal conversion of pyrazine (C4H4N2 D2h) is the best-known example of ultrafast electronic deactivation mediatedby a conical intersection The low-lying conical intersectionbetween the S2 and S1 potential energy surfaces of pyrazinewas identified in 198836 Thereafter a number of theoreticalstudies have been performed on this system37shy50 Although

pyrazine has 24 normal modes only a single mode Q10a (b1g)mediates the S2shyS1 coupling and a few totally symmetric (ag)modes play principal roles in the vibrational dynamics Seeland Domcke proposed TRPES of pyrazine theoretically in1991 using model calculations taking into account three vibra-tional coordinates of Q1 Q6a and Q10a51 Theoretical studieshave predicted that the internal conversion occurs within30 fs while the time resolution of our experiment in 1999was 450 fs2627 Thus we constructed a sub-20 fs DUV laserusing filamentation four-wave mixing and revisited the S2shyS1dynamics of pyrazine in 201052

Figure 4 shows the UV photoabsorption spectra of pyrazine(pyrazine-h4) and fully deuterated pyrazine (pyrazine-d4) vapormeasured at room temperature Overlaid are the spectra of our3frac12 (264 nm blue) and 4frac12 (198 nm red) pulses The 3frac12 pulseexcites pyrazine to S2 near the origin and the 4frac12 pulse ionizesfrom S2 and S1 However the 4frac12 pulse overlaps (unfavorably)

Figure 3 (a) Schematic diagram of the experimental setup (b) Schematic drawing of the differential pumping systemAll dimensions are given in millimeters

Figure 4 UV photoabsorption spectra of S1 S2 and S3 ofpyrazine-h4 (thin solid line) and pyrazine-d4 (thin dashedline) at room temperature The spectra of our pump(264 nm 470 eV) and probe (198 nm 626 eV) pulses arealso shown in solid lines

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)344

with the S3shyS0 band and the 4frac12shy3frac12 pulse sequence in thenegative time delays also creates a photoelectron signal

Figure 5a shows the total photoelectron signal intensity asa function of the pumpshyprobe time delay For positive timedelays with the 3frac12shy4frac12 pulse sequence the total signal inten-sity rapidly decays within the first 30 fs and exhibits a plateauat later times our previous experiment in 1999 has shown thatthis long-lived component has a finite lifetime of 22 ps forpyrazine-h42753 For negative time delays with the 4frac12shy3frac12pulse sequence the signal decays within 50 fs These timeprofiles therefore consist of three components correspondingto the decay of optically excited S2 (red) the correspondinggrowth of S1 (blue) populated by internal conversion from S2and the decay of S3 (green) respectively Using least-squaresfitting the S2 frac14 S1 internal conversion time constants areestimated as 23 laquo 4 fs for pyrazine-h4 and 20 laquo 2 fs forpyrazine-d4 The plateau region exhibits oscillatory featuresdue to vibrational quantum beats of Q6a in S1 (583 cmsup11)

Figure 5b shows the photoelectron kinetic energy distribu-tions measured at each time delay Although the S2 frac14 S1internal conversion occurs within 30 fs no marked change isobserved here This is because photoionization predominantly

occurs as D0(nsup11)larr S1(nπ) and D1(πsup11)larr S2(ππ) and theenergy differences are almost identical between D1 and D0

(088 eV)54 and between S2 and S1 (086 eV)55

Figure 5c shows the time and energy dependence of theanisotropy parameter cent2(tE) the positive (blue-green) andnegative (red) values correspond to preferential ejection of anelectron parallel and perpendicular to the probe laser polar-ization (eq 1) The energy-dependence of cent2 a stripe of colorsat each time delay in Figure 5c is the fingerprint of theelectronic character In the region of t lt 30 fs high-energyelectrons are ejected perpendicularly to the probe laser polar-ization (red) while low-energy electrons are distributed almostisotropically (green) After 30 fs the high-energy electrondistribution becomes isotropic while low-energy electrons arepreferentially ejected parallel to the probe laser polarization(blue) The sudden change of the color at ca 30 fs indicates theoccurrence of S2 frac14 S1 internal conversion in agreement withthe analysis of the total electron signal

In Figure 5c the photoelectron kinetic energy component of08shy10 eV appearing within 30 fs is attributed to photoioniza-tion from S2(ππ) to D0(nsup11) As shown in Figure 6 themain electron configurations of S2(ππ) to D0(nsup11) cannot beconnected by a one-electron process The occurrence of theD0(nsup11) larr S2(ππ) ionization suggests that there are otherelectron configurations involved in the cationic andor neutralstate Within the HartreeshyFock approximation the electronicconfiguration of pyrazine in its ground state (S0) is given by

1b21u1a2g1b

22u2a

2g2b

21u1b

23g3a

2g3b

21u2b

22u4a

2g2b

23g5a

2g3b

22u4b

21u4b

22u3b

23g

1b23uethsup3THORN5b21uethnTHORN1b22gethsup3THORN1b21gethsup3THORN6a2gethnTHORN2b03uethsup3THORN1a0uethsup3THORN2b02gethsup3THORNeth2THORN

where nonbonding and π orbitals are respectively denoted byn and π and all other orbitals are σ orbitals Denoting the differ-ence from eq 2 the leading configurations of D0 and S2 areexpressed by 6a1

g and 1b11g 2b

thorn13u respectively To enable D0 larr

S2 ionization D0 andor S2 must involve minor electron con-figurations with a singly occupied b2u orbital and two-electronexcitation from the leading configuration jb1

2u 1b11g 2b

thorn13u ( j =

1 2 3 and 4) and 6a1g jbthorn1

2u ethj 5THORN are the candidates forD0 and S2 respectively as shown in Figures 6b and 6d Inorder to ascertain this speculation we performed first-orderconfiguration interaction and continuum multiple scattering

Figure 5 (a) Temporal profile of total photoelectron signalin 3frac12shy4frac12 TRPEI of pyrazine-h4 The observed profile iswell explained by three components single-exponentialdecay of S2 (red) corresponding increase in S1 (blue) forpositive time delays and single-exponential decay of S3(green) for negative time delays The fitting result is shownas a solid line (b) Time-evolution of photoelectron kineticenergy (PKE) distribution middot(tE) (c) Time evolution of thephotoelectron angular anisotropy parameter cent2(tE)

D0(Ag)

S2(B2u)

2b3u

2b3u

6ag

6ag

1b1g

1b1g

j b2u

j b2u

i b2u

Figure 6 Main configurations of (a) D0 and (c) S2 andrelevant configurations for one-photon ionization of D0 larrS2 (b) and (d)

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 345

Xα (CMSXα) calculations which indicated that the spectralintensity of D0 larr S2 is 47 from 3b2u 27 from 4b2u and26 from all virtual b2u ( jb2u j sup2 5)56

The characteristic negative cent2 associated with the D0 larr S2transition assisted identification of the S2 frac14 S1 internal con-version in pyrazine Using CMSXα calculations we found thatcent parameter diminishes with the electron kinetic energy owingto the energy-dependent Coulomb phases but cent does notbecome negative without a shape resonance caused by a short-lived bound electronic state buried in the ionization continuumOur calculations indicate that there is a shape resonance at35shy40 eV above the ionization energy56 Aromatic moleculesoften have shape resonances within the range of excess energyless than 10 eV which should be taken into account in theinterpretation of TRPEI

Figure 5b indicates that the FranckshyCondon distributionupon photoionization extends toward an unobservable negativekinetic energy region A probe photon with higher energy willenable observation of the entire FranckshyCondon envelopeThus we revisited the same reaction using the filamentationVUV light source TRPES of pyrazine using the 3frac12 pump and5frac12 probe is shown in Figure 7 In the positive time delayregion photoionization from the S2(ππ) and S1(nπ) areclearly seen in the range of 15shy25 eV Notice that the S1component has shifted to higher photoelectron kinetic energiesin Figure 7 in comparison with Figure 5b Since 5frac12 is stronglyabsorbed by pyrazine the 5frac12shy3frac12 pulse sequence also producesa signal in the negative time delay region The three peaksin the energy range of 15shy27 eV are of the 3s 3pz and 3pyRydberg states5457 The distribution in the 0shy2 eV range is dueto photodissociation from higher valence states excited with5frac12 so that the photoelectron kinetic energy diminishes rapidlyas the vibrational wave packet moves away from the FranckshyCondon region

23 S2 frac14 S1 Internal Conversion via Conical Intersectionin Benzene and Toluene Benzene is a benchmark aromaticmolecule for theoretical and experimental studies of organiccompounds The S2(1B1u) states of benzene and its derivativesare short-lived owing to ultrafast S2 frac14 S1 internal conversion

Let us examine the dynamics in comparison with the internalconversion in pyrazine

Figures 8a and 8b show the 4frac12 pump and 3frac12 probe TRPESof benzene and toluene respectively58 The broken lines aresingle exponential decay functions convoluted with the cross-correlation of the pump and probe pulses The discrepancybetween the data points and the simulated curves indicate thatthe single exponential decay model does not adequately re-produce the observed time profiles in either case The observeddecay profiles exhibit delayed responses which correspond topropagation of a wave packet from the FranckshyCondon regionto the conical intersection region The arrival times of the wavepackets at the seam of crossings are estimated as 33 fs forbenzene and 41 fs for toluene the subsequent population decaytime is 32 fs for benzene and 43 fs for toluene

Figure 7 Photoelectron kinetic energy distribution ob-served using the 3frac12 and 5frac12 pulses Positive time delayscorrespond to the case where the 3frac12 pulse precedes the5frac12 pulse

Figure 8 The time profiles of the photoionization signalintensity for (a) benzene and (b) toluene Photoelectronsignals are indicated by green dots with error bars Timeshyenergy maps of the photoelectron intensity σ(tE) for(c) benzene and (d) toluene Timeshyenergy maps of thephotoelectron angular anisotropies cent2(tE) for (e) benzeneand (f ) toluene and cent4(tE) for (g) benzene and (h) tolueneData points for cent2 and cent4 with standard deviations smallerthan 02 are shown (see text) Reproduced with permissionfrom Ref 58

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)346

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

with the S3shyS0 band and the 4frac12shy3frac12 pulse sequence in thenegative time delays also creates a photoelectron signal

Figure 5a shows the total photoelectron signal intensity asa function of the pumpshyprobe time delay For positive timedelays with the 3frac12shy4frac12 pulse sequence the total signal inten-sity rapidly decays within the first 30 fs and exhibits a plateauat later times our previous experiment in 1999 has shown thatthis long-lived component has a finite lifetime of 22 ps forpyrazine-h42753 For negative time delays with the 4frac12shy3frac12pulse sequence the signal decays within 50 fs These timeprofiles therefore consist of three components correspondingto the decay of optically excited S2 (red) the correspondinggrowth of S1 (blue) populated by internal conversion from S2and the decay of S3 (green) respectively Using least-squaresfitting the S2 frac14 S1 internal conversion time constants areestimated as 23 laquo 4 fs for pyrazine-h4 and 20 laquo 2 fs forpyrazine-d4 The plateau region exhibits oscillatory featuresdue to vibrational quantum beats of Q6a in S1 (583 cmsup11)

Figure 5b shows the photoelectron kinetic energy distribu-tions measured at each time delay Although the S2 frac14 S1internal conversion occurs within 30 fs no marked change isobserved here This is because photoionization predominantly

occurs as D0(nsup11)larr S1(nπ) and D1(πsup11)larr S2(ππ) and theenergy differences are almost identical between D1 and D0

(088 eV)54 and between S2 and S1 (086 eV)55

Figure 5c shows the time and energy dependence of theanisotropy parameter cent2(tE) the positive (blue-green) andnegative (red) values correspond to preferential ejection of anelectron parallel and perpendicular to the probe laser polar-ization (eq 1) The energy-dependence of cent2 a stripe of colorsat each time delay in Figure 5c is the fingerprint of theelectronic character In the region of t lt 30 fs high-energyelectrons are ejected perpendicularly to the probe laser polar-ization (red) while low-energy electrons are distributed almostisotropically (green) After 30 fs the high-energy electrondistribution becomes isotropic while low-energy electrons arepreferentially ejected parallel to the probe laser polarization(blue) The sudden change of the color at ca 30 fs indicates theoccurrence of S2 frac14 S1 internal conversion in agreement withthe analysis of the total electron signal

In Figure 5c the photoelectron kinetic energy component of08shy10 eV appearing within 30 fs is attributed to photoioniza-tion from S2(ππ) to D0(nsup11) As shown in Figure 6 themain electron configurations of S2(ππ) to D0(nsup11) cannot beconnected by a one-electron process The occurrence of theD0(nsup11) larr S2(ππ) ionization suggests that there are otherelectron configurations involved in the cationic andor neutralstate Within the HartreeshyFock approximation the electronicconfiguration of pyrazine in its ground state (S0) is given by

1b21u1a2g1b

22u2a

2g2b

21u1b

23g3a

2g3b

21u2b

22u4a

2g2b

23g5a

2g3b

22u4b

21u4b

22u3b

23g

1b23uethsup3THORN5b21uethnTHORN1b22gethsup3THORN1b21gethsup3THORN6a2gethnTHORN2b03uethsup3THORN1a0uethsup3THORN2b02gethsup3THORNeth2THORN

where nonbonding and π orbitals are respectively denoted byn and π and all other orbitals are σ orbitals Denoting the differ-ence from eq 2 the leading configurations of D0 and S2 areexpressed by 6a1

g and 1b11g 2b

thorn13u respectively To enable D0 larr

S2 ionization D0 andor S2 must involve minor electron con-figurations with a singly occupied b2u orbital and two-electronexcitation from the leading configuration jb1

2u 1b11g 2b

thorn13u ( j =

1 2 3 and 4) and 6a1g jbthorn1

2u ethj 5THORN are the candidates forD0 and S2 respectively as shown in Figures 6b and 6d Inorder to ascertain this speculation we performed first-orderconfiguration interaction and continuum multiple scattering

Figure 5 (a) Temporal profile of total photoelectron signalin 3frac12shy4frac12 TRPEI of pyrazine-h4 The observed profile iswell explained by three components single-exponentialdecay of S2 (red) corresponding increase in S1 (blue) forpositive time delays and single-exponential decay of S3(green) for negative time delays The fitting result is shownas a solid line (b) Time-evolution of photoelectron kineticenergy (PKE) distribution middot(tE) (c) Time evolution of thephotoelectron angular anisotropy parameter cent2(tE)

D0(Ag)

S2(B2u)

2b3u

2b3u

6ag

6ag

1b1g

1b1g

j b2u

j b2u

i b2u

Figure 6 Main configurations of (a) D0 and (c) S2 andrelevant configurations for one-photon ionization of D0 larrS2 (b) and (d)

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 345

Xα (CMSXα) calculations which indicated that the spectralintensity of D0 larr S2 is 47 from 3b2u 27 from 4b2u and26 from all virtual b2u ( jb2u j sup2 5)56

The characteristic negative cent2 associated with the D0 larr S2transition assisted identification of the S2 frac14 S1 internal con-version in pyrazine Using CMSXα calculations we found thatcent parameter diminishes with the electron kinetic energy owingto the energy-dependent Coulomb phases but cent does notbecome negative without a shape resonance caused by a short-lived bound electronic state buried in the ionization continuumOur calculations indicate that there is a shape resonance at35shy40 eV above the ionization energy56 Aromatic moleculesoften have shape resonances within the range of excess energyless than 10 eV which should be taken into account in theinterpretation of TRPEI

Figure 5b indicates that the FranckshyCondon distributionupon photoionization extends toward an unobservable negativekinetic energy region A probe photon with higher energy willenable observation of the entire FranckshyCondon envelopeThus we revisited the same reaction using the filamentationVUV light source TRPES of pyrazine using the 3frac12 pump and5frac12 probe is shown in Figure 7 In the positive time delayregion photoionization from the S2(ππ) and S1(nπ) areclearly seen in the range of 15shy25 eV Notice that the S1component has shifted to higher photoelectron kinetic energiesin Figure 7 in comparison with Figure 5b Since 5frac12 is stronglyabsorbed by pyrazine the 5frac12shy3frac12 pulse sequence also producesa signal in the negative time delay region The three peaksin the energy range of 15shy27 eV are of the 3s 3pz and 3pyRydberg states5457 The distribution in the 0shy2 eV range is dueto photodissociation from higher valence states excited with5frac12 so that the photoelectron kinetic energy diminishes rapidlyas the vibrational wave packet moves away from the FranckshyCondon region

23 S2 frac14 S1 Internal Conversion via Conical Intersectionin Benzene and Toluene Benzene is a benchmark aromaticmolecule for theoretical and experimental studies of organiccompounds The S2(1B1u) states of benzene and its derivativesare short-lived owing to ultrafast S2 frac14 S1 internal conversion

Let us examine the dynamics in comparison with the internalconversion in pyrazine

Figures 8a and 8b show the 4frac12 pump and 3frac12 probe TRPESof benzene and toluene respectively58 The broken lines aresingle exponential decay functions convoluted with the cross-correlation of the pump and probe pulses The discrepancybetween the data points and the simulated curves indicate thatthe single exponential decay model does not adequately re-produce the observed time profiles in either case The observeddecay profiles exhibit delayed responses which correspond topropagation of a wave packet from the FranckshyCondon regionto the conical intersection region The arrival times of the wavepackets at the seam of crossings are estimated as 33 fs forbenzene and 41 fs for toluene the subsequent population decaytime is 32 fs for benzene and 43 fs for toluene

Figure 7 Photoelectron kinetic energy distribution ob-served using the 3frac12 and 5frac12 pulses Positive time delayscorrespond to the case where the 3frac12 pulse precedes the5frac12 pulse

Figure 8 The time profiles of the photoionization signalintensity for (a) benzene and (b) toluene Photoelectronsignals are indicated by green dots with error bars Timeshyenergy maps of the photoelectron intensity σ(tE) for(c) benzene and (d) toluene Timeshyenergy maps of thephotoelectron angular anisotropies cent2(tE) for (e) benzeneand (f ) toluene and cent4(tE) for (g) benzene and (h) tolueneData points for cent2 and cent4 with standard deviations smallerthan 02 are shown (see text) Reproduced with permissionfrom Ref 58

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)346

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

Xα (CMSXα) calculations which indicated that the spectralintensity of D0 larr S2 is 47 from 3b2u 27 from 4b2u and26 from all virtual b2u ( jb2u j sup2 5)56

The characteristic negative cent2 associated with the D0 larr S2transition assisted identification of the S2 frac14 S1 internal con-version in pyrazine Using CMSXα calculations we found thatcent parameter diminishes with the electron kinetic energy owingto the energy-dependent Coulomb phases but cent does notbecome negative without a shape resonance caused by a short-lived bound electronic state buried in the ionization continuumOur calculations indicate that there is a shape resonance at35shy40 eV above the ionization energy56 Aromatic moleculesoften have shape resonances within the range of excess energyless than 10 eV which should be taken into account in theinterpretation of TRPEI

Figure 5b indicates that the FranckshyCondon distributionupon photoionization extends toward an unobservable negativekinetic energy region A probe photon with higher energy willenable observation of the entire FranckshyCondon envelopeThus we revisited the same reaction using the filamentationVUV light source TRPES of pyrazine using the 3frac12 pump and5frac12 probe is shown in Figure 7 In the positive time delayregion photoionization from the S2(ππ) and S1(nπ) areclearly seen in the range of 15shy25 eV Notice that the S1component has shifted to higher photoelectron kinetic energiesin Figure 7 in comparison with Figure 5b Since 5frac12 is stronglyabsorbed by pyrazine the 5frac12shy3frac12 pulse sequence also producesa signal in the negative time delay region The three peaksin the energy range of 15shy27 eV are of the 3s 3pz and 3pyRydberg states5457 The distribution in the 0shy2 eV range is dueto photodissociation from higher valence states excited with5frac12 so that the photoelectron kinetic energy diminishes rapidlyas the vibrational wave packet moves away from the FranckshyCondon region

23 S2 frac14 S1 Internal Conversion via Conical Intersectionin Benzene and Toluene Benzene is a benchmark aromaticmolecule for theoretical and experimental studies of organiccompounds The S2(1B1u) states of benzene and its derivativesare short-lived owing to ultrafast S2 frac14 S1 internal conversion

Let us examine the dynamics in comparison with the internalconversion in pyrazine

Figures 8a and 8b show the 4frac12 pump and 3frac12 probe TRPESof benzene and toluene respectively58 The broken lines aresingle exponential decay functions convoluted with the cross-correlation of the pump and probe pulses The discrepancybetween the data points and the simulated curves indicate thatthe single exponential decay model does not adequately re-produce the observed time profiles in either case The observeddecay profiles exhibit delayed responses which correspond topropagation of a wave packet from the FranckshyCondon regionto the conical intersection region The arrival times of the wavepackets at the seam of crossings are estimated as 33 fs forbenzene and 41 fs for toluene the subsequent population decaytime is 32 fs for benzene and 43 fs for toluene

Figure 7 Photoelectron kinetic energy distribution ob-served using the 3frac12 and 5frac12 pulses Positive time delayscorrespond to the case where the 3frac12 pulse precedes the5frac12 pulse

Figure 8 The time profiles of the photoionization signalintensity for (a) benzene and (b) toluene Photoelectronsignals are indicated by green dots with error bars Timeshyenergy maps of the photoelectron intensity σ(tE) for(c) benzene and (d) toluene Timeshyenergy maps of thephotoelectron angular anisotropies cent2(tE) for (e) benzeneand (f ) toluene and cent4(tE) for (g) benzene and (h) tolueneData points for cent2 and cent4 with standard deviations smallerthan 02 are shown (see text) Reproduced with permissionfrom Ref 58

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)346

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

Figures 8c and 8d show the time evolution of photoelectronkinetic energy distributions extracted from a series of imagestaken at different time delays for benzene and toluene respec-tively The distributions consist of the S2 and S1 componentsthat are largely different each other The S2 component mainlyappears in the 05shy15 eV range while S1 lt 03 eV Ionizationfrom S2 and S1 occurs to the same cationic states in which thevibrational energies are nearly conserved owing to the FranckshyCondon principle upon ionization and similar potential energysurfaces between the neutrals and ions Since internal con-version transforms the S2shyS1 electronic energy difference intothe vibrational energy in S1 photoionization from S1 occurs tohighly vibrationally excited states of the cation leaving smallphotoelectron kinetic energies5960 For both benzene andtoluene close examination of the high-energy (1shy15 eV) regionof the S2 component reveals oscillatory vibrational wave packetdynamics for example in the case of benzene the maximumof the energy distribution appears at 14 eV at 0 fs 10 eV at20 fs and 13 eV at 30 fs

Similarly to the case of pyrazine the ultrashort S2 lifetimeimplies that photoexcited benzene easily accesses an S2S1conical intersection region Theoretical calculations predictedthat the minimum-energy S2S1 conical intersection point(a prefulvenic form) is close in energy and structure to theS2 potential minimum66162 The minimum however is at anonplanar structure (a boat form) that differs from the planarstructure of benzene (D6h) in S066162 Consequently a photo-excited benzene molecule rapidly deforms from a planar struc-ture in the FranckshyCondon region toward a nonplanar structurealong the steepest descent of the S2 potential energy surfaceand undergoes a nonadiabatic transition in the vicinity of theS2S1 seam of crossings This is a different feature from thepyrazine case in which the FranckshyCondon region is close tothe minimum energy conical intersection point at the planargeometry (Figure 9) The hot S1 benzene produced by S2 frac14 S1internal conversion is further funneled down to S0 via the S1S0conical intersection in lt10 ps596263

Close examination of Figure 8e for benzene reveals that cent2

varies with time most clearly around 07 eV cent2 is negative att = 0 and gradually increases with time to be positive around

30 fs Similar time dependence of cent2 is also seen in Figure 8ffor toluene for example at around 10 eV The evolution ofcent2(tE) indicates that the electronic character gradually changesalong the out-of-plane distortion Notice that the out-of-planedistortion eliminates the planar symmetry and consequentlydistinction between the σ and π orbitals Thus different elec-tronic states at planar geometry are mixed at nonplanar geom-etries the mixing is most likely with 1E1u and 1E2g

3 Restoration of Sharp Photoelectron Angular Distributionfrom Broad Angular Distribution Observed

for Rotating Molecules

Chemists consider reaction mechanisms having in mindthe electron orbitals of a molecule fixed in space howeverreal molecules are randomly oriented in gases and liquids andthis leads to inevitable directional averaging in physicalmeasurements It is an experimental challenge to overcomethis difficulty and ldquowatchrdquo electronic structures and dynamics inthe molecular frame In one-photon excitation of randomlyoriented molecules with a linearly polarized light photoexcitedmolecules are weakly aligned in space with their transitiondipole moments distributed by cos2ordm around the polarizationof light where ordm is the angle of the transition dipole andpolarization The 4th order Legendre term in eq 1 arises fromthe alignment effect However the alignment created by one-photon excitation is so weak that even if a photoelectronangular distribution has sharp structures in the molecular framethe observed distribution in the laboratory frame is considerablyblurred by the broad molecular axis distribution Althougheven the blurred distributions provide valuable insights intononadiabatic dynamics as discussed in the previous sectionI discuss in this section restoration of a sharp photoelectronangular distribution seen by an observer on the rotatingmolecule from the blurred distributions The key element forsuccessful restoration is accurate information about a point-spread function which projects a single point in the originalsharp image into a finite region in a transformed (blurred)image In the present case the point-spread function is givenby the axis alignment factor A(t) The molecular axis distribu-tion P(raquo) in the laboratory frame is expressed as P(traquo) =middot4sup3[1 + A(t)P2(cosraquo)] using A(t) The important advantageof one-photon excitation is that A(t) can be rigorously calculatedand reliable analysis can be performed

In quantum mechanics localization of a particle in a freespace is realized by superposition of multiple eigenfunctionswith different momenta however localization is nonstationaryand the distribution function is rapidly broadened owing tophase dispersion of different momentum components Like-wise sharp directional alignment or orientation of a molecule isassociated with a rotational wave function with a broad angularmomentum distribution In one-photon excitation of moleculesthe selection rule of brvbarJ = 0 laquo1 creates a superposition of onlya few rotational states and a broad molecular axis distribution(J is the rotational angular momentum quantum number) Thequantum mechanical interference of rotational wave functionscreates peculiar time dependence of the molecular axis align-ment factor A(t) Let us examine it for the S1(nπ) frac14 T1(nπ)intersystem crossing in pyrazine This process occurs in 110 pswhich is comparable to the rotational period 82 ps of this

Figure 9 Comparison of conical intersections in pyrazineand benzene The vertical arrows indicate photoexcitationand the horizontal axes are the reaction coordinates Thered and green correspond respectively to the diabatic S2and S1 states and blue indicates contribution of other zero-order electronic states

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 347

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

molecule286465 After exciting pyrazine to the zero vibrationallevel in S1 we observed two-photon ionization from the S1 andT1 states using the 400 nm probe pulse in which ionization wasenhanced by accidental resonance with the Rydberg states64

The photoelectron signal intensities of S1 and T1 in Figure 10bexhibit respectively a gradual decay and growth as a functionof time in which characteristic spikes due to A(t) appeared(The S1 state is probed via both 3s and 3p Rydberg states5457

which exhibit opposite sign of the rotational coherence spikes)Similar signatures of A(t) is also seen in the photoelectronangular anisotropies (represented by cent20cent00 in this case) inFigure 10c

We demonstrated restoration of molecular frame photoelec-tron angular distribution using [1 + 1curren] ionization of NO via theA(2shy+) state as an example A free electron wave in Coulombicinteractions with a diatomic cation is expressed using theelectron partial waves such as sσ pσ pπ and others as thestandard method We derived the formulae that relate middot(tE)cent2(tE) and cent4(tE) with A(t) and the amplitudes and phasesof partial waves66shy68 and we measured these observables forvarious time-delays and probe photon energies (Figure 11)

These measurements provided sufficient number of experimen-tal observables to determine 11 values of the amplitudes andphases We excited NO(2not12) to the A(2shy+) state with a 226 nmpulse and probed by one-photon ionization at various probewavelengths from 323 to 2425 nm to change the photoelectronkinetic energy from 005 to 133 eV6768 One technologicalchallenge was that the cent2 and cent4 parameters needed to bedetermined with accuracies of 001 In order to achieve thisaccuracy we developed the 1 kHz CMOS camera and centroid-ing calculations using field programmable gate arrays andintegrated each image over 106 laser shots31

The molecular frame photoelectron angular distributionsthus determined are shown in Figure 12 for two different probephoton polarization directions and photoelectron kinetic en-ergies6768 Photoionization with the polarization parallel to thebond axis induces photoemission on the oxygen side at lowenergy while photoemission to the nitrogen side increases withenergy Polarization perpendicular to the bond axis creates abroad photoelectron angular distribution near the thresholdwhile it sharpens distributions along the probe laser polar-ization at higher kinetic energies

4 Time-Resolved Photoelectron Spectroscopy of Liquids

41 TRPES at Ultralow Kinetic Energies It is highlyinteresting to explore solution chemistry using TRPES On theother hand a crucial quantity in photoelectron spectroscopy ofcondensed matter is the escape depth of an electron Althoughexcitation light has a large penetration depth into the bulkmaterial electrons created far from the surface are unlikely toescape into vacuum owing to inelastic scattering in the bulkThe inelastic scattering cross section takes the maximum valueat around 50shy100 eV and diminishes in both lower and higherenergy sides69shy74 For example Figure 13 shows the integralcross sections of the inelastic scattering of an electron withthe electron energy loss larger than 1 eV in amorphous ice75

The result indicates that the cross section diminishes by nearlythree orders of magnitudes for E lt 8 eV Thus one may expectthat the electron escape depth in liquid water increases at lowelectron kinetic energy However an electron undergoes exten-sive elastic scattering in liquid water at low kinetic energiesand the actual escape depth has not been determined yetExtensive studies on the escape depth of an electron fromliquid water is in progress in our laboratory a reasonableestimate is 2shy10 nm in the kinetic energy region less than 10 eVThis depth may seem rather short however the high relativepermittivity of water effectively screens electrostatic interac-tions by hydration making the liquid properties rapidlyapproach those of the bulk at small depths from the surface

42 Charge Transfer to Solvent Reaction from Isup1 to BulkWater In order to characterize TRPES of liquids we studiedthe charge transfer to solvent (CTTS) reaction from Isup1 to bulkwater which has previously been studied using conventionaltransient absorption spectroscopy7677 When Isup1(aq) is photo-excited to a metastable excited state below the conductionband an excess electron cannot be ejected freely into the bulkthe electron cloud can however penetrate into bulk water tobe trapped as a hydrated electron Internal conversion to theground state of Isup1(aq) competes with this CTTS reaction Whenaccommodating the excess electron a local hydrogen-bonding

Figure 10 (a) Two-dimensional slice image of photoelec-trons in the [1 + 2curren] PEI of pyrazine via the S1 B3u(nπ) 0deglevel observed at a time delay of 30 ps The original imagewas integrated for 80000 laser shots (b) Time evolution ofthe three major components in the [1 + 2curren] PEI Circles ( )triangles ( ) and squares ( ) represent the angle-integratedintensity for the outer (PKE = 643meV) middle (101meV) and inner rings (37meV) respectively Solidlines are a simulation taking into account the rotationalcoherence (c) cent20cent00 as a function of 5 defined in the textReproduced with permission from Ref 64

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)348

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

network has to change drastically by reorienting water mole-cules For the studies of CTTS dynamics strongly coupled withsolvation dynamics Isup1(aq) is a suitable solute as it has nointernal degrees of freedom besides electronic motion We

excited Isup1(aq) in water with a pump pulse (243shy226 nm) andphotodetached negatively charged transients with a time-delayed 260 nm probe pulse (Figure 14)19 Figure 15 showsthe photoelectron spectrum measured as a function of the time

Figure 11 (a) The excitation scheme from NO (X2not12 vPrime = 0) to NO+ (1shy+ v+ = 0) via the A (2shy+ vcurren = 0) state Panels (b) (c)and (d) show axis alignment factor A(t) and time energy maps of photoelectron intensity and anisotropy parameters measured asfunctions of photoelectron kinetic energy and pumpshyprobe time delay middot(tE) in (b) represents the normalized values at each PKEThe anisotropy parameters were measured at around 42 and 84 ps with a time window of 06 ps and the anisotropy parameters atother time delays were calculated from these experimental values to illustrate their time dependences Reproduced with permissionfrom Ref 67

Figure 12 Molecular frame photoelectron angular distributions reconstructed from timeshyenergy mapping of photoelectronanisotropy parameters for the σ (a) and π (b) channels Reproduced with permission from Ref 68

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 349

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

delay The energy spectrum is initially broad but the widthdiminishes rapidly and the intensity decreases gradually Ifelectron inelastic scattering in water is extensive the photo-electron kinetic energy distribution should exhibit a deceleratedcomponent irrespective of the pumpshyprobe time delay How-ever the result exhibits no signature of decelerated electronsindicating that the influence of inelastic scattering on theenergy distribution can be neglected

Figure 16 shows the total electron intensity observed as afunction of delay time for the CTTS reactions in H2O andD2O78 The decay profiles clearly indicate that there are atleast three elementary steps involved in the CTTS reactionand the time constants are different between the two solventsThe time-dependences of the photoelectron signal intensity andspectral shifts are explained by a model assuming two reactionintermediates of a solvent separated and contact pair states78

The observed decay is ascribed to geminate recombination of

an electron with a neutral iodine atom from the CTTS solventseparated and contact pair state The time constants extractedfrom TRPES (Figure 17) are in good agreement with thosefrom transient absorption spectroscopy7677 Thus although theprobing depth of TRPES is not established yet it is stronglysuggested that TRPES at ultralow kinetic energy (lt5 eV)probes bulk solution chemistry

43 Electron-Binding Energies of Solvated ElectronsThe CTTS reaction in bulk water discussed in the precedingsection ultimately creates hydrated electrons1979shy98 The hy-drated electron is the most important transient species inradiation chemistry and biology its electron-binding energyhowever has not been measured by any methods In the 1990s

Figure 13 Integral cross section (shown by dots) ascribedto the sum of dissociative attachment (DA) vibrationalexcitation above 1 eV energy loss electronic excitations(electr) as well as ionization (ion) processes in amorphousice Adopted from Ref 75

Figure 14 Schematic energy diagram for time-resolvedphotoelectron spectroscopy of a charge transfer to solvent(CTTS) reaction Reproduced with permission fromRef 78

Figure 15 3D plot of the photoelectron kinetic energy dis-tribution as a function of time in [1 + 1curren] TRPES of 014Maqueous NaI solution The pump and probe wavelengthsare 243 and 260 nm respectively Reproduced withpermission from Ref 19

100500

delay timeps

0

05

1

phot

oele

ctro

n si

gnal

(ar

b u

nit)

200

1

H2O

D2OH2O

D2O

Figure 16 Photoelectron signal intensity as a function ofpumpshyprobe time delay observed for the charge transfer tosolvent (CTTS) reaction of Isup1 to bulk water the samplesolutions were 01M aqueous NaI solution in H2O (black)and D2O (blue) The schematic energy diagram of theexperiment is shown in Figure 14 The cross correlation ofthe pump and probe laser pulses was ca 300 fs A hemi-spherical electron energy analyzer was used to measurea photoelectron spectrum at each time delay and thespectrum was integrated over photoelectron kinetic energyto obtain the each data point Reproduced with permissionfrom Ref 78

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)350

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

researchers found that the electron-binding energy of nega-tively charged water clusters varies approximately as n13

where n is the number of water molecules in the cluster Thistrend agrees qualitatively with a model of an electron in aspherical dielectric cavity In the 1990s the electron-bindingenergy of a hydrated electron in bulk water was estimated to be33 eV by extrapolating the vertical electron-binding energiesof negatively charged water clusters to the infinitely large sizeof the cluster87 Theoretical calculations however have refutedthis estimate by showing that the electron-binding energyvaries with n13 even when the excess electron is trapped atthe surfaces of clusters9099100 Since the cluster anions aregenerated by electron attachment to cold neutral clusters inmolecular beams the energetic penalty is large for an excesselectron to penetrate into a cluster by disrupting the hydrogen-bonding network consequently an electron is trapped at thesurface of cold clusters Later studies have shown that thereare three different isomers (IshyIII) for negatively charged waterclusters The isomers IshyIII have three different asymptoticvalues of electron-binding energies at the infinitely large sizesof the clusters in which isomer I is regarded as a precursor of ahydrated electron in the bulk101 while isomer II seems relatedto a hypothesized surface state of a hydrated electron102103

A more recent study has revealed that the isomer I has twodifferent species known as isomer Ia and Ib the extrapolationof the electron-binding energy to the infinite cluster size of Ib isin the region of 35shy40 eV (Figure 18)94104

Our first TRPES experiment of aqueous NaI solutionestimated the vertical electron-binding energy of a hydratedelectron in bulk water to be 33 eV (Figure 15)19 in excellentagreement with the estimate made from the cluster values8687

However a low signal-to-noise ratio and electrokinetic charg-ing of a liquid beam limited the accuracy In order to achievehigher precision and reliability for liquid TRPES we haveconstructed a magnetic bottle time-of-flight (TOF) spectrometer

(flight length 1m) which collects 50 of photoelectronsemitted from the liquid beam and enables measurement of theentire photoelectron energy spectrum on a shot-to-shot basis105

In combination with a 100 kHz DUV femtosecond laser weconstructed the new spectrometer improved the signal countlevel by four to five orders of magnitude in comparison withour old system The spectra of solvated electrons in H2O D2Omethanol and ethanol measured at a pumpshyprobe time delayof 2 ns are shown in Figure 19 A striking feature of thesespectra is their symmetric Gaussian shapes the photoabsorp-tion spectra of solvated electrons in bulk solutions as well asphotoelectron spectra of water cluster anions exhibit asym-

Figure 17 Graphical presentation of our kinetic modelwith some representative time constants for the chargetransfer to solvent (CTTS) reaction of Isup1 to bulk water Theobserved experimental results can be explained equallywell by a kinetic model presented here and a diffusionmodel The latter involves solving a diffusion equationwith an assumed interaction potential eg a Morse poten-tial Reproduced with permission from Ref 78

Figure 18 Vertical binding energy of a hydrated electron inbulk water observed in this work compared with the clustervalues by Ma et al104 The original extrapolation by Coeet al is also indicated by a broken line87 Reproduced withpermission from Ref 19

Figure 19 Photoelectron spectra of solvated electrons inbulk solutions H2O (black) D2O (blue) methanol (green)and ethanol (red) Open circles are experimental data andsolid lines are the best-fit Gaussian functions obtainedby least-squares fitting There was no signal in the energyregion of 0shy15 eV Reproduced with permission fromRef 105

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 351

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

metric band shapes which are fit by a Gaussian and LorentzianThe electron-binding energies of solvated electrons determinedfrom these high-precision measurements are listed in Table 1

As seen in Table 1 the electron-binding energy of a solvatedelectron in methanol is 34 eV which is clearly larger than thevalue 26 eV estimated from the methanol cluster data106

The result indicates that the excellent agreement betweenthe electron-binding energy of a hydrated electron in bulksolution and the estimate from water cluster anions cannot begeneralized to other solvents Molecular cluster anions pro-duced in supersonic molecular beams are at ultralow temper-atures and their thermodynamic properties are generally differ-ent from bulk solution at an ambient temperature Therefore itis possible that the agreement observed for water is fortuitous

5 Summary

Time-resolved photoelectron imaging (TRPEI) enablesefficient and accurate measurements of photoelectron energyand angular distributions as a function of the pumpshyprobe timedelay Timeshyenergy mapping of photoelectron angular aniso-tropy clearly reveals rapid changes of electron configurationduring the course of photoinduced dynamics The S2 stateof pyrazine exhibited immediate population decay from theFranckshyCondon region while the S2 state of benzene andtoluene decayed after finite induction time due to wave packetmotions from the FranckshyCondon region to conical intersectionregion The results illustrate the importance of the locationsof the conical intersections on the potential energy surfacesin internal conversion dynamics It has been shown that thephotoelectron angular anisotropy is influenced by shape res-onances with metastable bound states in the continuum

Time-resolved photoelectron spectroscopy (TRPES) ofliquids provides new opportunities for ultrafast spectroscopicstudies of electron dynamics under wet conditions TRPES ofthe charge transfer to solvent reaction from Isup1 to bulk waterwell demonstrated the usefulness of this methodology for thestudies of solution chemistry The electron-binding energiesof solvated electrons in four polar protic solvents were firmlyestablished by liquid TRPES The method is expected to beparticularly useful for elucidating electron-transfer processesand redox reactions Similarly with TRPES in the gas phasethe timeshyenergy mapping of photoelectron angular anisotropybecomes an invaluable tool in studying electronic dynamicsin solution and its technical development is underway in ourlaboratory

I would like to thank my present and past group memberswho made great contribution to TRPES in our laboratory YSuzuki T Fuji Y Tang T Horio H Kohguchi Y Ogi S-Y

Liu H Shen K Sekiguchi and N Kurahashi I also thankProfessors V Bonacic-Koutechy and R Mitiric for collabo-ration on molecular dynamics on the fly

References

1 M Born R Oppenheimer Ann Phys 1927 389 4572 E Teller Isr J Chem 1969 7 2273 L Salem J Am Chem Soc 1974 96 34864 L Salem C Leforestier G Segal R Wetmore J Am

Chem Soc 1975 97 4795 J Michl V Bonacic-Koutecky Electronic Aspects of

Organic Photochemistry Wiley-Interscience Publication 19906 I J Palmer I N Ragazos F Bernardi M Olivucci M A

Robb J Am Chem Soc 1993 115 6737 C Nordling E Sokolowski K Siegbahn Phys Rev 1957

105 16768 D W Turner M I Al Jobory J Chem Phys 1962 37

30079 F I Vilesov A N Terenin B L Kurbatoy Dokl Akad

Nauk SSSR 1961 138 132910 K Siegbahn J Electron Spectrosc Relat Phenom 1985

36 11311 J L Knee in High Resolution Laser Photoionization and

Photoelectron Studies ed by I Powis T Baer C-Y Ng JohnWiley amp Sons 199512 I V Hertel W Radloff Rep Prog Phys 2006 69 189713 T Seideman Annu Rev Phys Chem 2002 53 4114 K L Reid Int Rev Phys Chem 2008 27 60715 A Stolow A E Bragg D M Neumark Chem Rev 2004

104 171916 H Siegbahn J Phys Chem 1985 89 89717 M Faubel S Schlemmer J P Toennies Z Phys D At

Mol Clusters 1988 10 26918 K R Wilson B S Rude J Smith C Cappa D T Co

R D Schaller M Larsson T Catalano R J Saykally Rev SciInstrum 2004 75 72519 Y Tang H Shen K Sekiguchi N Kurahashi T Mizuno

Y-I Suzuki T Suzuki Phys Chem Chem Phys 2010 12 365320 Y Tang Y-i Suzuki H Shen K Sekiguchi N Kurahashi

K Nishizawa P Zuo T Suzuki Chem Phys Lett 2010 494 11121 P Kruit F H Read J Phys E Sci Instrum 1983 16 31322 D W Chandler P L Houston J Chem Phys 1987 87

144523 A T J B Eppink D H Parker Rev Sci Instrum 1997

68 347724 J J Lin J Zhou W Shiu K Liu Rev Sci Instrum 2003

74 249525 S-Y Liu K Alnama J Matsumoto K Nishizawa H

Kohguchi Y-P Lee T Suzuki J Phys Chem A 2011 115 295326 T Suzuki L Wang H Kohguchi J Chem Phys 1999

111 485927 L Wang H Kohguchi T Suzuki Faraday Discuss 1999

113 3728 T Suzuki in Modern Trends in Chemical Reaction

Dynamics Experiment and Theory (Part I) ed by C-Y NgWorld Scientific Singapore 200429 T Suzuki Annu Rev Phys Chem 2006 57 55530 T Suzuki Int Rev Phys Chem 2012 31 26531 T Horio T Suzuki Rev Sci Instrum 2009 80 01370632 T Fuji T Horio T Suzuki Opt Lett 2007 32 248133 T Fuji T Suzuki E E Serebryannikov A Zheltikov

Table 1 Vertical Binding Energies (eV) of SolvatedElectrons

Solvent Liquid Estimate from cluster anion

H2O 34 33 (Ia)a) 40 (Ib)b)

D2O 35 reg

Methanol 34 26c)

Ethanol 33 reg

a) Ref 87 b) Ref 104 c) Ref 106

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)352

Phys Rev A 2009 80 06382234 P Zuo T Fuji T Horio S Adachi T Suzuki Appl Phys

B 2012 108 81535 S L Chin Femtosecond Laser Filamentation in Springer

Series on Atomic Optical and Plasma Physics Springer NewYork 2010 doi101007978-1-4419-0688-536 R Schneider W Domcke Chem Phys Lett 1988 150

23537 L Seidner G Stock A L Sobolewski W Domcke

J Chem Phys 1992 96 529838 C Woywod W Domcke A L Sobolewski H-J Werner

J Chem Phys 1994 100 140039 T Gerdts U Manthe Chem Phys Lett 1998 295 16740 A Raab G A Worth H-D Meyer L S Cederbaum

J Chem Phys 1999 110 93641 M Thoss W H Miller G Stock J Chem Phys 2000

112 1028242 C Coletti G D Billing Chem Phys Lett 2003 368 28943 D V Shalashilin M S Child J Chem Phys 2004 121

356344 X Chen V S Batista J Chem Phys 2006 125 12431345 P Puzari B Sarkar S Adhikari J Chem Phys 2006 125

19431646 P Puzari R S Swathi B Sarkar S Adhikari J Chem

Phys 2005 123 13431747 R He C Zhu C-H Chin S H Lin Chem Phys Lett

2009 476 1948 U Werner R Mitrić T Suzuki V Bonačić-Kouteckyacute

Chem Phys 2008 349 31949 U Werner R Mitrić V Bonačić-Kouteckyacute J Chem Phys

2010 132 17430150 C-K Lin Y Niu C Zhu Z Shuai S H Lin Chemreg

Asian J 2011 6 297751 M Seel W Domcke J Chem Phys 1991 95 780652 Y-I Suzuki T Fuji T Horio T Suzuki J Chem Phys

2010 132 17430253 V Stert P Farmanara W Radloff J Chem Phys 2000

112 446054 M Oku Y Hou X Xing B Reed H Xu C Chang C-Y

Ng K Nishizawa K Ohshimo T Suzuki J Phys Chem A 2008112 229355 K K Innes I G Ross W R Moomaw J Mol Spectrosc

1988 132 49256 Y-I Suzuki T Suzuki J Chem Phys 2012 137 19431457 J K Song M Tsubouchi T Suzuki J Chem Phys 2001

115 881058 Y-I Suzuki T Horio T Fuji T Suzuki J Chem Phys

2011 134 18431359 W Radloff V Stert Th Freudenberg I V Hertel C

Jouvet C Dedonder-Lardeux D Solgadi Chem Phys Lett 1997281 2060 P Farmanara V Stert W Radloff I V Hertel J Phys

Chem A 2001 105 561361 M Meisl R Janoschek J Chem Soc Chem Commun

1986 106662 A Toniolo A L Thompson T J Martiacutenez Chem Phys

2004 304 13363 W Radloff T Freudenberg H-H Ritze V Stert F Noack

I V Hertel Chem Phys Lett 1996 261 30164 M Tsubouchi B J Whitaker L Wang H Kohguchi T

Suzuki Phys Rev Lett 2001 86 450065 T Suzuki L Wang M Tsubouchi J Phys Chem A 2004

108 576466 Y-I Suzuki T Suzuki Mol Phys 2007 105 167567 Y Tang Y-I Suzuki T Horio T Suzuki Phys Rev Lett

2010 104 07300268 Y-I Suzuki Y Tang T Suzuki Phys Chem Chem Phys

2012 14 730969 M P Seah W A Dench Surf Interface Anal 1979 1 270 T L Barr Modern ESCA The Principles and Practice of

X-Ray Photoelectron Spectroscopy CRC Press Inc 199471 S Huumlfner Photoelectron Spectroscopy Principles and

Applications in Advanced Texts in Physics Springer-Verlag 2003doi101007978-3-662-09280-472 D Emfietzoglou I Kyriakou I Abril R Garcia-Molina

H Nikjoo Int J Radiat Biol 2012 88 2273 D Emfietzoglou I Kyriakou I Abril R Garcia-Molina

I D Petsalakis H Nikjoo A Pathak Nucl Instrum MethodsPhys Res Sect B 2009 267 4574 N Ottosson M Faubel S E Bradforth P Jungwirth

B Winter J Electron Spectrosc Relat Phenom 2010 177 6075 M Michaud A Wen L Sanche Radiat Res 2003 159 376 J A Kloepfer V H Vilchiz V A Lenchenkov X Chen

S E Bradforth J Chem Phys 2002 117 76677 H Iglev A Trifonov A Thaller I Buchvarov T Fiebig

A Laubereau Chem Phys Lett 2005 403 19878 Y-I Suzuki H Shen Y Tang N Kurahashi K Sekiguchi

T Mizuno T Suzuki Chem Sci 2011 2 109479 E J Hart J W Boag J Am Chem Soc 1962 84 409080 M Natori T Watanabe J Phys Soc Jpn 1966 21 157381 S Schlick P A Narayana L Kevan J Chem Phys 1976

64 315382 F J Webster J Schnitker M S Friedrichs R A Friesner

P J Rossky Phys Rev Lett 1991 66 317283 J Schnitker K Motakabbir P J Rossky R A Friesner

Phys Rev Lett 1988 60 45684 P J Rossky J Schnitker J Phys Chem 1988 92 427785 J Schnitker P J Rossky J Chem Phys 1987 86 347186 A E Bragg J R R Verlet A Kammrath O

Cheshnovsky D M Neumark Science 2004 306 66987 J V Coe G H Lee J G Eaton S T Arnold H W

Sarkas K H Bowen C Ludewigt H Haberland D R WorsnopJ Chem Phys 1990 92 398088 M J Tauber R A Mathies J Phys Chem A 2001 105

1095289 M Mizuno T Tahara J Phys Chem A 2001 105 882390 L Turi W-S Sheu P J Rossky Science 2005 309 91491 I A Shkrob J Phys Chem A 2007 111 522392 I A Shkrob W J Glover R E Larsen B J Schwartz

J Phys Chem A 2007 111 523293 R E Larsen W J Glover B J Schwartz Science 2010

329 6594 L D Jacobson J M Herbert J Am Chem Soc 2011

133 1988995 J M Herbert L D Jacobson Int Rev Phys Chem 2011

30 196 K R Siefermann Y Liu E Lugovoy O Link M Faubel

U Buck B Winter B Abel Nat Chem 2010 2 27497 A T Shreve T A Yen D M Neumark Chem Phys Lett

2010 493 21698 A Luumlbcke F Buchner N Heine I V Hertel T Schultz

Phys Chem Chem Phys 2010 12 1462999 O Marsalek F Uhlig T Frigato B Schmidt P Jungwirth

Phys Rev Lett 2010 105 043002

T Suzuki Bull Chem Soc Jpn Vol 87 No 3 (2014) 353

100 O Marsalek F Uhlig P Jungwirth J Phys Chem C 2010114 20489101 J V Coe S M Williams K H Bowen Int Rev PhysChem 2008 27 27102 J R R Verlet A E Bragg A Kammrath OCheshnovsky D M Neumark Science 2005 307 93103 J R R Verlet A E Bragg A Kammrath O

Cheshnovsky D M Neumark Science 2005 310 1769104 L Ma K Majer F Chirot B von Issendorff J ChemPhys 2009 131 144303105 T Horio H Shen S Adachi T Suzuki Chem Phys Lett2012 535 12106 A Kammrath J R R Verlet G B Griffin D MNeumark J Chem Phys 2006 125 171102

Toshinori Suzuki obtained his PhD degree in 1988 from Tohoku University He then became aresearch associate at the Institute for Molecular Science (IMS) between 1988 and 1990 and aJSPS fellow for research abroad to carry out research on molecular beam scattering at CornellUniversity and the University of California Berkeley between 1990 and 1992 He returned toJapan as an associate professor at IMS where he started his independent research group onchemical reaction dynamics in 1992 In 2001 he moved to RIKEN to be a chief scientist andextended his research further In 2009 he became a professor of physical chemistry at KyotoUniversity to educate the next leaders of Japanese science and from 2012 he is serving as a Chairof the Division of Chemistry Graduate School of Science He received Broida Award IBMScience Award JSPS Award Commendation for Science and Technology by MEXT

AWARD ACCOUNTSBull Chem Soc Jpn Vol 87 No 3 (2014)354